summaryrefslogtreecommitdiff
path: root/Build/source/texk/web2c/uptexdir/kanji.c
diff options
context:
space:
mode:
authorTakuji Tanaka <ttk@t-lab.opal.ne.jp>2019-02-23 01:59:36 +0000
committerTakuji Tanaka <ttk@t-lab.opal.ne.jp>2019-02-23 01:59:36 +0000
commit0527b2f34a785f177c6038093da7a5ab4edfb76b (patch)
treed3667cd8a82dd06235ba61796853c5719d0532a1 /Build/source/texk/web2c/uptexdir/kanji.c
parentbcbdd79cfec3e2407f9ff948a53747672f05e8ff (diff)
upTeX 1.24
git-svn-id: svn://tug.org/texlive/trunk@50095 c570f23f-e606-0410-a88d-b1316a301751
Diffstat (limited to 'Build/source/texk/web2c/uptexdir/kanji.c')
-rw-r--r--Build/source/texk/web2c/uptexdir/kanji.c168
1 files changed, 94 insertions, 74 deletions
diff --git a/Build/source/texk/web2c/uptexdir/kanji.c b/Build/source/texk/web2c/uptexdir/kanji.c
index a96f7b9bbae..ae6d7b87013 100644
--- a/Build/source/texk/web2c/uptexdir/kanji.c
+++ b/Build/source/texk/web2c/uptexdir/kanji.c
@@ -27,7 +27,7 @@ boolean is_char_ascii(integer c)
boolean is_char_kanji(integer c)
{
if (is_internalUPTEX())
- return (c>=0&&(c & CJK_TOKEN_FLAG) < CJK_CHAR_LIMIT);
+ return (c >= 0);
else
return iskanji1(Hi(c)) && iskanji2(Lo(c));
}
@@ -56,25 +56,25 @@ integer calc_pos(integer c)
}
/* Ref. http://www.unicode.org/Public/UNIDATA/Blocks.txt */
-/* # Blocks-10.0.0.txt */
-/* # Date: 2017-04-12, 17:30:00 GMT [KW] */
+/* # Blocks-12.0.0.txt */
+/* # Date: 2018-07-30, 19:40:00 GMT [KW] */
static long ucs_range[]={
0x0000, /* Basic Latin */ /* 0x00 */
0x0080, /* Latin-1 Supplement */
0x0100, /* Latin Extended-A */
0x0180, /* Latin Extended-B */
- 0x0250, /* IPA Extensions */
+ 0x0250, /* IPA Extensions */
0x02B0, /* Spacing Modifier Letters */
0x0300, /* Combining Diacritical Marks */
0x0370, /* Greek and Coptic */
0x0400, /* Cyrillic */
0x0500, /* Cyrillic Supplement */
0x0530, /* Armenian */
- 0x0590, /* Hebrew */
- 0x0600, /* Arabic */
- 0x0700, /* Syriac */
+ 0x0590, /* Hebrew */
+ 0x0600, /* Arabic */
+ 0x0700, /* Syriac */
0x0750, /* Arabic Supplement */
- 0x0780, /* Thaana */
+ 0x0780, /* Thaana */
0x07C0, /* NKo */ /* 0x10 */
0x0800, /* Samaritan */
0x0840, /* Mandaic */
@@ -86,7 +86,7 @@ static long ucs_range[]={
0x0A80, /* Gujarati */
0x0B00, /* Oriya */
0x0B80, /* Tamil */
- 0x0C00, /* Telugu */
+ 0x0C00, /* Telugu */
0x0C80, /* Kannada */
0x0D00, /* Malayalam */
0x0D80, /* Sinhala */
@@ -108,9 +108,9 @@ static long ucs_range[]={
0x1760, /* Tagbanwa */
0x1780, /* Khmer */
0x1800, /* Mongolian */ /* 0x30 */
- 0x18B0, /* Unified Canadian Aboriginal Syllabics Extended */
+ 0x18B0, /* Unified Canadian Aboriginal Syllabics Extended */
0x1900, /* Limbu */
- 0x1950, /* Tai Le */
+ 0x1950, /* Tai Le */
0x1980, /* New Tai Lue */
0x19E0, /* Khmer Symbols */
0x1A00, /* Buginese */
@@ -119,30 +119,31 @@ static long ucs_range[]={
0x1B00, /* Balinese */
0x1B80, /* Sundanese */
0x1BC0, /* Batak */
- 0x1C00, /* Lepcha */
+ 0x1C00, /* Lepcha */
0x1C50, /* Ol Chiki */
0x1C80, /* Cyrillic Extended-C */
- 0x1CC0, /* Sundanese Supplement */
- 0x1CD0, /* Vedic Extensions */ /* 0x40 */
+ 0x1C90, /* Georgian Extended */
+ 0x1CC0, /* Sundanese Supplement */ /* 0x40 */
+ 0x1CD0, /* Vedic Extensions */
0x1D00, /* Phonetic Extensions */
- 0x1D80, /* Phonetic Extensions Supplement */
- 0x1DC0, /* Combining Diacritical Marks Supplement */
+ 0x1D80, /* Phonetic Extensions Supplement */
+ 0x1DC0, /* Combining Diacritical Marks Supplement */
0x1E00, /* Latin Extended Additional */
- 0x1F00, /* Greek Extended */
+ 0x1F00, /* Greek Extended */
0x2000, /* General Punctuation */
0x2070, /* Superscripts and Subscripts */
0x20A0, /* Currency Symbols */
0x20D0, /* Combining Diacritical Marks for Symbols */
0x2100, /* Letterlike Symbols */
0x2150, /* Number Forms */
- 0x2190, /* Arrows */
- 0x2200, /* Mathematical Operators */
+ 0x2190, /* Arrows */
+ 0x2200, /* Mathematical Operators */
0x2300, /* Miscellaneous Technical */
- 0x2400, /* Control Pictures */
- 0x2440, /* Optical Character Recognition */ /* 0x50 */
- 0x2460, /* Enclosed Alphanumerics */
+ 0x2400, /* Control Pictures */ /* 0x50 */
+ 0x2440, /* Optical Character Recognition */
+ 0x2460, /* Enclosed Alphanumerics */
0x2500, /* Box Drawing */
- 0x2580, /* Block Elements */
+ 0x2580, /* Block Elements */
0x25A0, /* Geometric Shapes */
0x2600, /* Miscellaneous Symbols */
0x2700, /* Dingbats */
@@ -154,8 +155,8 @@ static long ucs_range[]={
0x2A00, /* Supplemental Mathematical Operators */
0x2B00, /* Miscellaneous Symbols and Arrows */
0x2C00, /* Glagolitic */
- 0x2C60, /* Latin Extended-C */
- 0x2C80, /* Coptic */ /* 0x60 */
+ 0x2C60, /* Latin Extended-C */ /* 0x60 */
+ 0x2C80, /* Coptic */
0x2D00, /* Georgian Supplement */
0x2D30, /* Tifinagh */
0x2D80, /* Ethiopic Extended */
@@ -169,15 +170,15 @@ static long ucs_range[]={
0x30A0, /* Katakana */
0x3100, /* Bopomofo */
0x3130, /* Hangul Compatibility Jamo */
- 0x3190, /* Kanbun */
- 0x31A0, /* Bopomofo Extended */
- 0x31C0, /* CJK Strokes */ /* 0x70 */
+ 0x3190, /* Kanbun */
+ 0x31A0, /* Bopomofo Extended */ /* 0x70 */
+ 0x31C0, /* CJK Strokes */
0x31F0, /* Katakana Phonetic Extensions */
0x3200, /* Enclosed CJK Letters and Months */
0x3300, /* CJK Compatibility */
0x3400, /* CJK Unified Ideographs Extension A */
0x4DC0, /* Yijing Hexagram Symbols */
- 0x4E00, /* CJK Unified Ideographs */
+ 0x4E00, /* CJK Unified Ideographs */
0xA000, /* Yi Syllables */
0xA490, /* Yi Radicals */
0xA4D0, /* Lisu */
@@ -186,14 +187,14 @@ static long ucs_range[]={
0xA6A0, /* Bamum */
0xA700, /* Modifier Tone Letters */
0xA720, /* Latin Extended-D */
- 0xA800, /* Syloti Nagri */
- 0xA830, /* Common Indic Number Forms */ /* 0x80 */
+ 0xA800, /* Syloti Nagri */ /* 0x80 */
+ 0xA830, /* Common Indic Number Forms */
0xA840, /* Phags-pa */
0xA880, /* Saurashtra */
0xA8E0, /* Devanagari Extended */
0xA900, /* Kayah Li */
- 0xA930, /* Rejang */
- 0xA960, /* Hangul Jamo Extended-A */
+ 0xA930, /* Rejang */
+ 0xA960, /* Hangul Jamo Extended-A */
0xA980, /* Javanese */
0xA9E0, /* Myanmar Extended-B */
0xAA00, /* Cham */
@@ -202,31 +203,31 @@ static long ucs_range[]={
0xAAE0, /* Meetei Mayek Extensions */
0xAB00, /* Ethiopic Extended-A */
0xAB30, /* Latin Extended-E */
- 0xAB70, /* Cherokee Supplement */
- 0xABC0, /* Meetei Mayek */ /* 0x90 */
+ 0xAB70, /* Cherokee Supplement */ /* 0x90 */
+ 0xABC0, /* Meetei Mayek */
0xAC00, /* Hangul Syllables */
- 0xD7B0, /* Hangul Jamo Extended-B */
+ 0xD7B0, /* Hangul Jamo Extended-B */
0xD800, /* High Surrogates */
0xDB80, /* High Private Use Surrogates */
- 0xDC00, /* Low Surrogates */
+ 0xDC00, /* Low Surrogates */
0xE000, /* Private Use Area */
0xF900, /* CJK Compatibility Ideographs */
0xFB00, /* Alphabetic Presentation Forms */
0xFB50, /* Arabic Presentation Forms-A */
0xFE00, /* Variation Selectors */
- 0xFE10, /* Vertical Forms */
+ 0xFE10, /* Vertical Forms */
0xFE20, /* Combining Half Marks */
0xFE30, /* CJK Compatibility Forms */
0xFE50, /* Small Form Variants */
- 0xFE70, /* Arabic Presentation Forms-B */
- 0xFF00, /* Halfwidth and Fullwidth Forms */ /* 0xa0 */
- 0xFFF0, /* Specials */
+ 0xFE70, /* Arabic Presentation Forms-B */ /* 0xa0 */
+ 0xFF00, /* Halfwidth and Fullwidth Forms */
+ 0xFFF0, /* Specials */
0x10000, /* Linear B Syllabary */
0x10080, /* Linear B Ideograms */
0x10100, /* Aegean Numbers */
- 0x10140, /* Ancient Greek Numbers */
+ 0x10140, /* Ancient Greek Numbers */
0x10190, /* Ancient Symbols */
- 0x101D0, /* Phaistos Disc */
+ 0x101D0, /* Phaistos Disc */
0x10280, /* Lycian */
0x102A0, /* Carian */
0x102E0, /* Coptic Epact Numbers */
@@ -234,11 +235,11 @@ static long ucs_range[]={
0x10330, /* Gothic */
0x10350, /* Old Permic */
0x10380, /* Ugaritic */
- 0x103A0, /* Old Persian */
- 0x10400, /* Deseret */ /* 0xb0 */
+ 0x103A0, /* Old Persian */ /* 0xb0 */
+ 0x10400, /* Deseret */
0x10450, /* Shavian */
0x10480, /* Osmanya */
- 0x104B0, /* Osage */
+ 0x104B0, /* Osage */
0x10500, /* Elbasan */
0x10530, /* Caucasian Albanian */
0x10600, /* Linear A */
@@ -250,24 +251,28 @@ static long ucs_range[]={
0x10900, /* Phoenician */
0x10920, /* Lydian */
0x10980, /* Meroitic Hieroglyphs */
- 0x109A0, /* Meroitic Cursive */
- 0x10A00, /* Kharoshthi */ /* 0xc0 */
+ 0x109A0, /* Meroitic Cursive */ /* 0xc0 */
+ 0x10A00, /* Kharoshthi */
0x10A60, /* Old South Arabian */
0x10A80, /* Old North Arabian */
0x10AC0, /* Manichaean */
0x10B00, /* Avestan */
0x10B40, /* Inscriptional Parthian */
- 0x10B60, /* Inscriptional Pahlavi */
+ 0x10B60, /* Inscriptional Pahlavi */
0x10B80, /* Psalter Pahlavi */
0x10C00, /* Old Turkic */
- 0x10C80, /* Old Hungarian */
+ 0x10C80, /* Old Hungarian */
+ 0x10D00, /* Hanifi Rohingya */
0x10E60, /* Rumi Numeral Symbols */
- 0x11000, /* Brahmi */
+ 0x10F00, /* Old Sogdian */
+ 0x10F30, /* Sogdian */
+ 0x10FE0, /* Elymaic */
+ 0x11000, /* Brahmi */ /* 0xd0 */
0x11080, /* Kaithi */
0x110D0, /* Sora Sompeng */
0x11100, /* Chakma */
0x11150, /* Mahajani */
- 0x11180, /* Sharada */ /* 0xd0 */
+ 0x11180, /* Sharada */
0x111E0, /* Sinhala Archaic Numbers */
0x11200, /* Khojki */
0x11280, /* Multani */
@@ -278,60 +283,75 @@ static long ucs_range[]={
0x11580, /* Siddham */
0x11600, /* Modi */
0x11660, /* Mongolian Supplement */
- 0x11680, /* Takri */
+ 0x11680, /* Takri */ /* 0xe0 */
0x11700, /* Ahom */
+ 0x11800, /* Dogra */
0x118A0, /* Warang Citi */
+ 0x119A0, /* Nandinagari */
0x11A00, /* Zanabazar Square */
0x11A50, /* Soyombo */
- 0x11AC0, /* Pau Cin Hau */ /* 0xe0 */
+ 0x11AC0, /* Pau Cin Hau */
0x11C00, /* Bhaiksuki */
0x11C70, /* Marchen */
- 0x11D00, /* Masaram Gondi */
+ 0x11D00, /* Masaram Gondi */
+ 0x11D60, /* Gunjala Gondi */
+ 0x11EE0, /* Makasar */
+ 0x11FC0, /* Tamil Supplement */
0x12000, /* Cuneiform */
0x12400, /* Cuneiform Numbers and Punctuation */
- 0x12480, /* Early Dynastic Cuneiform */
+ 0x12480, /* Early Dynastic Cuneiform */ /* 0xf0 */
0x13000, /* Egyptian Hieroglyphs */
- 0x14400, /* Anatolian Hieroglyphs */
+ 0x13430, /* Egyptian Hieroglyph Format Controls */
+ 0x14400, /* Anatolian Hieroglyphs */
0x16800, /* Bamum Supplement */
0x16A40, /* Mro */
0x16AD0, /* Bassa Vah */
0x16B00, /* Pahawh Hmong */
+ 0x16E40, /* Medefaidrin */
0x16F00, /* Miao */
0x16FE0, /* Ideographic Symbols and Punctuation */
0x17000, /* Tangut */
- 0x18800, /* Tangut Components */ /* 0xf0 */
+ 0x18800, /* Tangut Components */
0x1B000, /* Kana Supplement */
0x1B100, /* Kana Extended-A */
- 0x1B170, /* Nushu */
+ 0x1B130, /* Small Kana Extension */
+ 0x1B170, /* Nushu */ /* 0x100 */
0x1BC00, /* Duployan */
0x1BCA0, /* Shorthand Format Controls */
0x1D000, /* Byzantine Musical Symbols */
0x1D100, /* Musical Symbols */
0x1D200, /* Ancient Greek Musical Notation */
- 0x1D300, /* Tai Xuan Jing Symbols */
- 0x1D360, /* Counting Rod Numerals */
+ 0x1D2E0, /* Mayan Numerals */
+ 0x1D300, /* Tai Xuan Jing Symbols */
+ 0x1D360, /* Counting Rod Numerals */
0x1D400, /* Mathematical Alphanumeric Symbols */
0x1D800, /* Sutton SignWriting */
- 0x1E000, /* Glagolitic Supplement */
- 0x1E800, /* Mende Kikakui */
- 0x1E900, /* Adlam */
- 0x1EE00, /* Arabic Mathematical Alphabetic Symbols */ /* 0x100 */
- 0x1F000, /* Mahjong Tiles */
+ 0x1E000, /* Glagolitic Supplement */
+ 0x1E100, /* Nyiakeng Puachue Hmong */
+ 0x1E2C0, /* Wancho */
+ 0x1E800, /* Mende Kikakui */
+ 0x1E900, /* Adlam */
+ 0x1EC70, /* Indic Siyaq Numbers */ /* 0x110 */
+ 0x1ED00, /* Ottoman Siyaq Numbers */
+ 0x1EE00, /* Arabic Mathematical Alphabetic Symbols */
+ 0x1F000, /* Mahjong Tiles */
0x1F030, /* Domino Tiles */
- 0x1F0A0, /* Playing Cards */
+ 0x1F0A0, /* Playing Cards */
0x1F100, /* Enclosed Alphanumeric Supplement */
0x1F200, /* Enclosed Ideographic Supplement */
- 0x1F300, /* Miscellaneous Symbols and Pictographs */
+ 0x1F300, /* Miscellaneous Symbols and Pictographs */
0x1F600, /* Emoticons */
0x1F650, /* Ornamental Dingbats */
0x1F680, /* Transport and Map Symbols */
0x1F700, /* Alchemical Symbols */
0x1F780, /* Geometric Shapes Extended */
- 0x1F800, /* Supplemental Arrows-C */
+ 0x1F800, /* Supplemental Arrows-C */
0x1F900, /* Supplemental Symbols and Pictographs */
+ 0x1FA00, /* Chess Symbols */ /* 0x120 */
+ 0x1FA70, /* Symbols and Pictographs Extended-A */
0x20000, /* CJK Unified Ideographs Extension B */
0x2A700, /* CJK Unified Ideographs Extension C */
- 0x2B740, /* CJK Unified Ideographs Extension D */ /* 0x110 */
+ 0x2B740, /* CJK Unified Ideographs Extension D */
0x2B820, /* CJK Unified Ideographs Extension E */
0x2CEB0, /* CJK Unified Ideographs Extension F */
0x2F800, /* CJK Compatibility Ideographs Supplement */
@@ -343,11 +363,11 @@ static long ucs_range[]={
0x80000, /* reserved */
0x90000, /* reserved */
0xA0000, /* reserved */
- 0xB0000, /* reserved */
+ 0xB0000, /* reserved */ /* 0x130 */
0xC0000, /* reserved */
0xD0000, /* reserved */
0xE0000, /* Tags */
- 0xE0100, /* Variation Selectors Supplement */ /* 0x120 */
+ 0xE0100, /* Variation Selectors Supplement */
0xF0000, /* Supplementary Private Use Area-A */
0x100000, /* Supplementary Private Use Area-B */
/* Value over 0x10FFFF is illegal under Unicode,
@@ -361,11 +381,11 @@ static long ucs_range[]={
0x170000, /* Reserved */
0x180000, /* Reserved */
0x190000, /* Reserved */
- 0x1A0000, /* Reserved */
+ 0x1A0000, /* Reserved */ /* 0x140 */
0x1B0000, /* Reserved */
0x1C0000, /* Reserved */
0x1D0000, /* Reserved */
- 0x1E0000, /* Reserved */ /* 0x130 */
+ 0x1E0000, /* Reserved */
0x1F0000, /* Reserved */
0x200000, /* Reserved */
0x210000, /* Reserved */